Individual variation in face processing
Lead Research Organisation:
University of Glasgow
Department Name: School of Psychology
Abstract
Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.
Organisations
Publications
Burton AM
(2010)
The Glasgow Face Matching Test.
in Behavior research methods
Megreya AM
(2011)
The other-race effect does not rely on memory: Evidence from a matching task.
in Quarterly journal of experimental psychology (2006)
| Description | Some people are better at recognising faces than others. This is a commonplace observation, and yet it has been virtually ignored by the large body of research into human face recognition. In this project, we explored the variation in face recognition abilities, for three reasons. First, we wished to quantify the range of abilities in the population, across a variety of different tests (e.g., face memory, face matching, processing familiar and unfamiliar faces). Second, we wanted to use these differences to explore unresolved theoretical issues in face processing, specifically whether all tasks with faces involve shared mental processes, or whether these are to some extent independent. Third, we appreciate that accurate face recognition is increasingly important to professionals, such as police officers, or those checking security ID. We wished to establish whether these professionals have superior face recognition abilities, by virtue of their experience, and whether any such expertise could be captured for future selection or training. In the first part of the project, we established a new database of faces, for use in research. We captured images of over 300 volunteers, using three cameras (two still cameras, and a digital video camera). The resulting database (the Glasgow Unfamiliar Face Database, GUFD) comprises very many images of the same person, taken in different poses on the same day. We believe this will become a significant resource for the face research community. We used this database to construct a new test (the Glasgow Face Matching Test, GFMT) in which observers are shown pairs of faces, and asked to decide whether these belong to the same or different people. We used this in a larger battery of tests including: unfamiliar face matching and memory tasks, familiar face matching and memory tasks, and object matching and memory tasks. The battery was applied to a large cohort of 300 general public volunteers. The results of this large-scale investigation revealed substantial individual variations in all aspects of face processing. We have made this normative data available to the research community. The results also show that performance on tests with objects tend to correlate with performance on unfamiliar, but not familiar faces. This is of considerable importance, since it suggests that processing changes qualitatively as a face becomes familiar. Infollow-uptests,werecruitedparticipantswhohadperformedpoorlyonthe face matching tests, and administered a further batch of tests. We established that these people performed poorly at almost all face recognition tasks, not simply those with unfamiliar face matching. However, they showed completely normal levels of emotion recognition. This is particularly interesting, since it suggests that the processes involved in identity and emotion perception are to some extent independent. Interestingly, even though our poor face recognisers were consistently below the population average on 7 REFERENCE No. many tests, they showed no awareness of this when questioned about their abilities. In a second strand of the project, we examined experts involved in face processing professionally. Our main cohort of participants was police officers, and throughout the project we tested 150 serving officers, and a further 194 police probationers. These were very widely varying in their seniority and experience, from constables through to senior detectives. However, all had had standard police training in identification. We used a number of different face tests, including unfamiliar face matching, to compare the performance of police officers to that of students (who are the typical sample in academic face research). Across very many types of test, the main finding was that police officers perform no better than students. Despite this, the confidence of police officers in their performance is normally reliably higher than that of students. This dissociation between perception and performance is clearly of considerable importance in the legal process. During the project, the opportunity arose to look at the 'other race effect'. It is well- established that people find it easier to remember faces of their own race than those of another race. However, it was not clear whether this advantage for own-race faces would extend to matching (i.e. are these two photos the same person?). In collaboration with a colleague in Egypt, we tested British and Egyptian participants, on British and Egyptian faces. Theresultsshowedaclearotherraceeffect.Eventhoughtimewasunlimited, and viewers did not have to remember any faces, participants found it easier to match two faces of their own race, than of the other race. We further tested this in British police officers. These officers showed exactly the same pattern as British students, i.e. they were more accurate with British than with Egyptian faces. |
| Exploitation Route | Finally, we were also able to test a number of other expert face processors. Although our sample numbers were small, we established that a group of passport officers were no better on face recognition tests than any of the other groups (students or police officers). Similarly, a group of CCTV operators also scored at the same levels. However, interestingly, a small group of experts from a forensic facial imaging bureau, used by police to establish issues of identity, performed better than any other participants studied across the whole project. In follow-up studies we attempted to establish the basis of this expert performance, which appears to be based on fine-grain analysis of facial areas not normally used in recognizing an identity (e.g. the ears). Our early attempts to teach this strategy to non-experts did not result in improved performance. However, the fact that these forensic experts can perform so well does demonstrate that the task is possible, and leads clearly to future research which will attempt to operationalise their procedures. |
| Sectors | Security and Diplomacy |
| Description | This research laid the foundation for an acceptance of large individual differences in face recognition ability. This is now used in contemporary methods of selection for security-critical jobs in which face recognition is key (police and passport officers). |
| First Year Of Impact | 2010 |
| Sector | Security and Diplomacy |
| Impact Types | Cultural |